Reaction rates for different minerals are usually measured in ideal conditions in batch experiments, where the impact of pore morphology and hydrodynamics have been fully neglected. Such reaction rates are used at continuum-scale (Darcy-scale) models without the impact of pore structure on upscaled reaction rates under flow conditions. Therefore, to address the gap from batch experiments to upscaled reaction rates in continuum-scale models, a pore-network model coupled with geochemical modeling has been developed. As a case study, we simulate the geochemical reactions of geothermal energy storage/recovery in sandstone rocks by coupling PhreeqcRM (a geochemistry model) with a pore-network model. The main purpose is to delineate the impact of pore morphology and dynamic conditions on upscaling of reaction rates using the surface-weighted and volume-weighted averaging. The results show that the kaolinite reaction rate in porous media highly depends on both the flow rate and spatial distribution of reactive pores. We evaluate the impact of correlation between the reactive pores and pore size distribution on upscaled reaction rates. Results indicate that if reactive pores do not belong to the main flow path, then upscaling the geochemical reactions based on the continuum-scale or batch experiments would be erroneous. In such a scenario, the discrepancy between volume-averaged and surface-weighted average reaction rates are highlighted. Moreover, increasing the injection flow rate results in lower average concentration of different species in the effluent, while it results in higher reaction rates in porous media. This research provides insights into the complex aspects of flow-based reaction rates versus the batch reaction rates. That has a significant impact on continuum-scale modeling of reactive transport for applications such as geothermal energy and enhanced oil recovery.